Middleware Scheduling Optimization Techniques for Distributed Real-Time and Embedded Systems

Developers of mission-critical distributed real-time and embedded (DRE) systems face a fundamental tension between (1) the performance gains achievable with hand-crafted optimizations to systems built largely from scratch and (2) the development cost and cycle-time reductions offered by common offthe-shelf (COTS) middleware, such as CORBA, COM+, and Java RMI. Current-generation COTS DRE middleware solutions, such as Real-Time CORBA 1.0 and the Real-Time Specification for Java (RTSJ), provide only low-level primitives for CPU management, and defer specific configuration and optimization decisions to system developers. This paper describes how the Kokyu portable middleware scheduling framework, which is built using standards-based COTS middleware and OS primitives, can be used both (1) to maintain the flexibility and re-use offered by COTS middleware and (2) to provide opportunities for domain-specific optimizations to meet stringent real-time performance requirements. We also present specific scheduling optimizations we are applying in collaboration with The Boeing Company to a research avionics mission computing platform, and describe extensions to our middleware scheduling framework to support these optimizations.